In the paper, we concentrate on one of the most challenging issues of economic geography, that is, how to explain and analyze the evolution of the spatial pattern of new industries. First, we review the literature concerning this issue. Various theoretical frameworks provide an explanation for the spatial evolution of new industries. One view is that the spatial ordering of industry is preordained by geographical endowments, transport possibilities and firms' needs. Weberian location theory, for example, assumes that new technologies develop most rapidly in those regions where their static, pre-given locational needs are most consistent with existing local factors. Another view criticizes such an idea of a preordained spatial pattern. According to the 'window of locational opportunity' view, it overlooks the possibility that (new) needs of new high-tech industries are not there from the outset but come into being as these industries develop. New high-tech firms require new resources (such as knowledge, skills, institutions) that are unlikely to be available in space. The firms have to create their own mechanisms to satisfy their needs and turn 'generic' resources (e.g. basic knowledge and skills) into 'specific' ones (e.g. specialized knowledge and skills). This approach leaves room for the existence of historical accidents in the spatial formation of new industries. New high tech firms are assumed to select their first location based on 'generic' resources and these are expected to be widely available in space. Finally, we discuss a few analytical issues that take up the question whether the spatial evolution of new industries should be described in such terms as necessity or chance. We take the example of the new Information and Communication Technology industry (ICT), which is now shaping the spatial system in Western economies. The depth and significance of the revolution in ICT is widely acknowledged. But how does the spatial pattern of this new industry develop? And how to determine analytically whether ...